The study, led by Ali Ala of TU Dublin and Saveetha Institute of Medical and Technical Sciences, together with Behnam Malmir, Hamed Baziyad, and Vladimir Simic, frames blockchain adoption not as a simple technology selection problem but as a cognitive reasoning challenge. Supply chain managers must judge interdependent criteria under uncertainty, weighing incomplete information and conflicting priorities. To capture that mental process, the researchers collected linguistic judgments from a panel of twenty experts—academics and industry practitioners, most with more than a decade of supply chain experience—and converted their assessments into triangular fuzzy numbers, a mathematical device that preserves the vagueness inherent in human language rather than forcing it into artificial precision.

The Fuzzy-DEMATEL method, an extension of the Decision-Making Trial and Evaluation Laboratory technique, then transforms those judgments into a total-relations matrix that captures both direct and indirect influences among criteria. From this matrix, the team computed two key statistics for each criterion: R, the total influence it exerts on the others, and C, the total influence it receives. The difference, R minus C, separates net causes from net effects, while the sum, R plus C, measures overall prominence within the system. The result is an impact relationship map—a visual, explainable picture of how experts believe the adoption ecosystem is wired.

The headline finding concerns smart contracts: self-executing, code-driven agreements that trigger transactions and certifications without intermediaries. Smart contracts recorded the highest outgoing influence of any criterion, at R = 3.46, and the strongest net causal effect, R minus C = +0.78. In plain terms, the experts judged that nearly every other capability in a blockchain deployment depends on contract logic. Smart contracts determine when a traceability record is written, under what conditions a payment settles, and what happens when a condition fails. Intervening on smart contracts changes what the system does; intervening on almost anything else changes only how much of it is visible.

Food traceability, by contrast, emerged as the most prominent criterion overall, with a combined influence score of R + C = 6.40, yet its net causality was nearly neutral at +0.06. The authors interpret this as the signature of a hub rather than a driver. Traceability simultaneously enables fraud detection and waste identification downstream while relying on smart contracts to generate records, secure storage to retain them, and payment events to populate them. It is the integration point of the system, not its origin. Alongside smart contracts, the analysis classified payment transactions, fraud prevention, and permanent secure storage as net causes, while food waste prevention, carbon emission detection, and power consumption fell firmly into the net-effect camp.

That last classification carries perhaps the most consequential practical implication. Food waste prevention and carbon emission detection sent no influence edges at or above the analytical threshold, marking them as pure receivers in the causal network. The authors argue this reflects a fundamental distinction between capabilities and measured outcomes: a blockchain deployment executes contracts, records provenance, and settles payments—things it does—whereas waste reduction and verified emissions are things it reveals or achieves. An outcome cannot be a precondition for the mechanism that produces it. For companies tempted to launch their first blockchain pilot around a sustainability story, the message is sobering: such pilots are likely to disappoint for reasons unrelated to the technology itself, because the capabilities they depend on have not yet been built.

The study’s robustness checks strengthen its claims considerably. Kendall’s coefficient of concordance showed strong agreement among the twenty experts (W = 0.72, p < 0.001), and a matrix-convergence test confirmed the stability of the aggregated judgments. Sensitivity analyses varied both the edge threshold and the prominence cutoff used to classify criteria; the cause-effect partition held across every setting for seven of the eight criteria. The lone exception was fraud prevention, which showed a slightly negative net causality at the lower bound of the fuzzy uncertainty range. Rather than dismissing this as noise, the authors read it as substantively meaningful: fraud prevention only operates where records are already trustworthy, so its causal power is conditional on prior infrastructure. They describe it as a conditional cause—high priority, but contingent on deployment maturity.

The findings also carry theoretical weight. The criteria were drawn from the Technology-Organization-Environment framework, transaction cost theory, and diffusion of innovation theory, yet the causal map revealed something the standard TOE formulation does not predict: technology-dimension criteria predominantly send influence while environment-dimension criteria predominantly receive it. If this pattern holds across contexts, the authors suggest, the three TOE dimensions may be organized sequentially rather than in parallel—a hypothesis they explicitly offer for future testing rather than a settled conclusion. Transaction cost theory, meanwhile, helps explain why smart contracts, traceability, and fraud prevention dominate: all three address the opportunistic behavior and information asymmetry that plague multi-party food chains.

For practitioners, the study proposes a three-stage implementation sequence. First comes the transactional foundation: define contract logic, data-writing rights, validation rules, and the consensus mechanism before any pilot begins—a stage at which energy concerns can also be largely resolved, since modern permissioned, Byzantine-fault-tolerant systems avoid the power-hungry proof-of-work designs of earlier blockchain generations. Second, once contract execution is reliable across at least two organizational boundaries, organizations should build operational capability in traceability, fraud prevention, and secure storage. Third, only after those layers exist should sustainability outcomes such as waste reduction and emissions verification be pursued. For policymakers, the analysis suggests that mandating sustainability disclosure before traceability infrastructure exists produces reports that cannot be independently verified; regulatory sequencing should mirror the implementation sequence, with data-format and interoperability standards first and verified sustainability reporting second.

The authors are careful about the limits of their evidence. The results derive from expert-elicited perceptions rather than longitudinal field performance data, from a single purposive panel, and from a criterion set deliberately restricted to eight items, excluding factors such as governance arrangements and interoperability with legacy systems. DEMATEL yields perceived directional relationships, not statistically tested causal effects. Still, the convergence between the expert-derived map and real-world deployment studies—including traceability platforms, halal-certification blockchains, and IoT-integrated smart-contract frameworks—lends external plausibility to the prioritization. Future work, the team notes, should test the proposed sequence with field data, larger and more diverse panels, and cross-country comparisons, and should extend the framework to the AI, IoT, and remote-sensing technologies increasingly layered atop distributed ledgers. For now, the study offers food supply chains something they have lacked: an explainable, evidence-based answer to the question of what to build first.

Subject of Research: Cognitive causal modeling of blockchain adoption criteria in agri-food supply chains using Fuzzy-DEMATEL

Article Title: Cognitive Causal Modeling of Blockchain Adoption in Agri-Food Supply Chains: A Fuzzy-DEMATEL Approach

Article References: Ala, A., Malmir, B., Baziyad, H., & Simic, V. (2026). Cognitive Causal Modeling of Blockchain Adoption in Agri-Food Supply Chains: A Fuzzy-DEMATEL Approach. Cognitive Computation, 18(1), Article 114. https://doi.org/10.1007/s12559-026-10660-0

Image Credits: AI Generated

DOI: 10.1007/s12559-026-10660-0

Keywords: blockchain, agri-food supply chain, Fuzzy-DEMATEL, smart contracts, food traceability, multi-criteria decision-making, sustainability, fraud prevention, cognitive computation, expert judgment, impact relationship map, technology adoption

The research, conducted by Zhongtang Guo of Duke Kunshan University, addresses a long-standing gap in EEG-based affective computing. Most existing studies of brain-driven emotion recognition have focused on classifying a person’s immediate emotional state at a given moment. Far less attention has been paid to predicting the summary judgment a person forms after an entire emotional sequence has unfolded. Yet it is precisely this retrospective evaluation that shapes clinical symptom recall, customer satisfaction, and everyday judgments of happiness or distress. The new work proposes that the neural signatures of the peak-end effect can be captured in event-related potentials, or ERPs, the millisecond-scale electrical ripples the brain produces in response to each stimulus.

To gather the necessary data, thirty right-handed participants aged 18 to 28 completed a sequential emotion induction paradigm built from the International Affective Picture System. Each trial presented six images in a row, arranged along a preset arousal curve so that each sequence contained a clear emotional peak and a defined endpoint. The researchers systematically manipulated where the peak occurred within the sequence and how intense the final image was, using a Latin square design to cross these variables fully. After each sequence, participants rated their overall experience on a nine-point scale, and these ratings became the training targets for the predictive model. Brain activity was recorded from a 64-channel EEG cap at a sampling rate of 1000 Hz, with rigorous preprocessing including independent component analysis to remove ocular, muscular, and cardiac artifacts.

Three ERP components formed the neural backbone of the system. The Early Posterior Negativity, appearing roughly 150 to 350 milliseconds after stimulus onset, indexes early attentional capture by emotionally salient material. The P300, measured between 300 and 500 milliseconds over central-parietal electrodes, reflects cognitive evaluation and stimulus categorization. The Late Positive Potential, or LPP, spanning roughly 400 to 800 milliseconds over midline parietal sites, is widely regarded as the core electrophysiological marker of emotional arousal and motivational salience. The study confirmed that all three components were significantly modulated by arousal intensity, with high-arousal negative images producing the largest LPP amplitudes, and the peak of each sequence was operationally defined as the image evoking the maximum absolute LPP amplitude.

On top of these neural features, the researchers built a hybrid deep learning architecture called TCN-Attention-PeakEnd Gate, abbreviated TAPE. The model begins with a temporal convolutional network, or TCN, whose dilated causal convolutions expand the receptive field exponentially across layers, allowing the network to capture long-range temporal dependencies without the vanishing-gradient problems of recurrent networks. The TCN output then passes through an eight-head multi-head self-attention module, which models global relationships among all time steps in the sequence. The distinctive element, however, is the peak-end gating module. This module takes the ERP feature vectors of the peak and endpoint moments and computes a learnable sigmoid gating signal that adaptively reweights the contribution of each time step. In effect, the peak-end rule is embedded into the network as a differentiable operation, constraining the model to mirror the cognitive bias that human memory actually exhibits.

The performance results were striking. Under leave-one-subject-out cross-validation, the most demanding evaluation scheme in EEG research, in which each participant serves as the test subject exactly once, TAPE achieved a mean absolute error of 1.038 rating points, a Pearson correlation of 0.654 between predicted and actual retrospective ratings, and a three-level classification accuracy of 70.4 percent. These figures significantly outperformed eight baselines spanning shallow regression, classical EEG feature engineering, and deep architectures including LSTM, CNN-LSTM, and Transformer models, with all differences confirmed by Holm-Bonferroni-corrected paired-sample t-tests. Across fifty independent fivefold cross-validation evaluations, TAPE simultaneously attained the highest median correlation of 0.682 and the smallest interquartile range of 0.054, indicating that its advantage was stable rather than an artifact of favorable data splits.

External validation on the publicly available SEED dataset replicated the model’s superiority, where TAPE reached 84.7 percent three-class accuracy compared with 81.9 percent for the strongest deep-learning baseline, a Transformer. An ablation study then dissected the contribution of each module. Removing the peak-end gating module increased error by roughly 11 percent, removing self-attention reduced the correlation by nearly 10 percent, and replacing the TCN encoder with an LSTM cost 6.3 percentage points of accuracy, confirming that each component makes an independent and synergistic contribution to performance.

Perhaps the most scientifically compelling findings came from the interpretability analysis. When the researchers visualized the attention weights learned by the trained model, they discovered that under high-arousal conditions the model automatically assigned its highest attention to the sequence positions where peak stimuli most frequently occurred, precisely as peak-end theory predicts. Under neutral conditions, where no salient peak existed, attention became nearly uniform across positions, consistent with the theoretical corollary that retrospective evaluation reverts to an averaging strategy when arousal variability is low. The model also revealed a previously underexplored valence asymmetry: endpoint attention weights were higher for positive sequences than negative ones, hinting at a neural basis for the everyday wisdom of ending experiences on a high note. Correlation analysis further showed that the gating signals tracked LPP amplitude more strongly than the other ERP components, with a correlation of 0.483, providing computational-level evidence that sustained emotional elaboration carried by the LPP is the primary neural signal underlying peak-end integration.

The implications reach well beyond the laboratory. In clinical psychology, retrospective symptom assessments such as the PHQ-9 depression questionnaire are known to be vulnerable to peak-end recall bias, and a theory-constrained decoding system of this kind could eventually support wearable-EEG monitoring that quantifies and corrects such distortions in depression, anxiety, and post-traumatic stress. In user experience research, combining EEG with peak-end gating offers a more objective way to identify the key moments that dominate a customer’s lasting impression than traditional questionnaires ever could. The authors also outline a roadmap for future work, including multi-center validation on larger and more diverse cohorts, real-time implementation on edge devices through model compression, multimodal integration with heart rate, skin conductance, and eye-tracking, and longitudinal clinical translation. The study’s broader methodological message may prove its most durable contribution: embedding mature cognitive theories directly into neural network architectures as differentiable constraints can simultaneously improve predictive accuracy, stability, and interpretability, offering a template for a new generation of theory-guided affective computing systems.

Subject of Research: A deep learning framework that predicts retrospective evaluations of emotional experiences from ERP neural markers of the peak-end effect

Article Title: A Deep Learning-Based Retrospective Evaluation Prediction System for Emotional Experiences: Temporal Dynamic Feature Extraction and ERP Neural Mechanisms of the Peak-End Effect

Article References: Guo, Z. (2026). A Deep Learning-Based Retrospective Evaluation Prediction System for Emotional Experiences: Temporal Dynamic Feature Extraction and ERP Neural Mechanisms of the Peak-End Effect. Cognitive Computation, 18(1), Article 111. https://doi.org/10.1007/s12559-026-10657-9

Image Credits: AI Generated

DOI: 10.1007/s12559-026-10657-9

Keywords: peak-end effect, deep learning, EEG, event-related potentials, emotion recognition, temporal convolutional network, self-attention, LPP, retrospective evaluation, affective computing, Cognitive Computation, brain-computer interface